DNA methylation has profound effects on the mammalian genome and is implicated in the inactivation of multiple tumor suppressor genes. Considerable variability exists in the number of methylated loci found in different cancers. We hypothesize that a classification of pancreatic carcinomas based on overall and specific gene methylation patterns will identify subgroups of cancers with a distinct clinicopathological features and with novel therapeutic targets. We have reported that a subset of pancreatic cancers (termed CIMP+ or CpG island methylation phenotype+) harbor a significantly higher prevalence of aberrant DNA methylation at endogenous tumor suppressor loci. We will classify pancreatic cancers based on their DNA methylation patterns and determine the significance of these patterns. The functional basis for the aberrant cancer DNA methylation patterns remain unknown. We hypothesize that aberrant methylation of some promoters may occur during carcinogenesis as a result of changes in their transcriptional activity. We will test this hypothesis by analyzing the methylation status of the promoters of genes transcriptionally activated or repressed by activation of the TGFB/SMAD4 pathway, a pathway commonly inactivated by mutation in pancreatic and other cancers. We also hypothesize that CIMP+ pancreatic cancers retain the normal de novo methylation function that is lost in CIMP- cancers. We hypothesize that some promoters become aberrantly methylated during carcinogenesis as a result of changes in their transcriptional activity, mediated by chromatin alterations and recruitment of DNA methyltransferases. This aims of this proposal have been identified by the program review group (PRG) of the NCI as a research priority-"identify genetic factors, environmental factors, and gene-environment interactions that contribute to development of pancreatic cancer." The clinicopathological, genetic and functional correlates of the CIMP classification of pancreatic cancer will be gained by the following specific aims: Aim #1 : Test the validity and significance of classifying pancreatic cancers based on methylator phenotypes. Aim #2: Determine the significance of aberrant methylation of specific genes in pancreatic cancer. Aim #3: Test whether genetic alterations in pancreatic carcinoma contribute to aberrant methylation in pancreatic cancer. Aim #4: Determine if functional DNA methylation differences exist between CIMP+ compared to CIMP- pancreatic carcinomas.